Alibaba has announced the open-source release of the DreaMoving framework, which enables ideal control of human motion based on diffusion models. This framework allows users to generate highly customized human video content, including scenarios such as a smiling girl on the beach or an Asian girl dancing in Central Park. By incorporating the Video ControlNet and Content Guider components, precise control over motion and appearance is achieved. Users only need to provide a portrait image and simple prompts to generate corresponding videos, with support for changing prompts, character backgrounds, and outfits. DreaMoving demonstrates strong generalization capabilities in the field of AI video generation, capable of producing high-quality videos based on guidance sequences and simple descriptions.